On scheduling two classes of real-time traffic with identical deadlines

The problem of scheduling two classes of real-time traffic with correlated time constraints is considered. Three scheduling disciplines are studied: a priority discipline which gives strict priority to one class of traffic, a threshold-based scheme in which priority is given to one class of traffic when the minimum laxity of its queued packets falls below some threshold, and a balancing scheme which assigns priority on the basis of the differences in minimum laxities in the two classes of traffic. Analytic results are obtained by using a discrete time model to obtain the state occupancy probabilities for the system. In this case, the state is defined using the laxities of the queued real-time packets. Parameters are defined to study the tradeoff in the performance of the two classes of traffic. Results are obtained to demonstrate how the balancing scheme makes it possible to achieve significant improvement in the performance of one class of traffic with only minimal effect on the performance of other classes. A video application is suggested for this work.<
>